From 137061be78bd97d8c775906fe372e8f136880c0f Mon Sep 17 00:00:00 2001
From: Dobrica Pavlinusic <dpavlin@rot13.org>
Date: Tue, 14 Nov 2023 10:00:04 +0100
Subject: [PATCH] similarity, forward progress only

---
 upari.pl | 33 +++++++++++----------------------
 1 file changed, 11 insertions(+), 22 deletions(-)

diff --git a/upari.pl b/upari.pl
index cf47bcd..4b09a69 100755
--- a/upari.pl
+++ b/upari.pl
@@ -1,5 +1,7 @@
 #!/usr/bin/perl
 
+# TODO: skola <-> razred swap?
+
 use warnings;
 use strict;
 use autodie;
@@ -29,9 +31,7 @@ foreach my $nr ( 1 .. 4 ) {
 			$row->[2],
 		);
 
-		my $num = $row->[0]; #$num =~ s/\D//g;
-
-		#$stat->{num}->{len}->{ length($num) }++;
+		my $num = uc $row->[0];
 
 		my $key_id = $row->[1] . '-' . $row->[2]; # . ',' . int( $row->[4] );
 
@@ -51,28 +51,19 @@ foreach my $nr ( 1 .. 4 ) {
 			}
 			if ( $#candidates == 0 ) {
 				$stat->{similarity}->{$nr}++;
-				warn "# similarity $num = ",dump( @candidates ),$/;
-				$id = $candidates[0];
+				$id = join('-',uc $candidates[0]->{key}, $row->[1], $row->[2]);
+				warn "# similarity $nr $num -> $id\n";
+				push @{ $keys->{ $key_id }->{ $id } }, uc $id;
 			} elsif ( $#candidates > 1 ) {
 				warn "# similarity IGNORED $num = ",dump( @candidates ),$/;
+				$stat->{similarity_multiple}->{$nr}++;
+			} else {
+				$stat->{similarity_none}->{$nr}++;
+				push @{ $keys->{ $key_id }->{ $num } }, uc $row->[0];
 			}
 
 
 
-			if ( exists $keys->{ $key_id }->{ $num } ) {
-			       	my @found = @{ $keys->{ $key_id }->{ $num } };
-				if ( $#found == 0 ) {
-					$id = $found[0];
-					print "MAPPED $nr $row->[0] to $id\n";
-					$stat->{mapped}->{$nr}++;
-				} elsif ( $#found > 0 ) {
-					print "ALIAS $nr $num into ",dump( @found ),$/;
-					$stat->{alias}->{$nr}++;
-				}
-			} else {
-				$stat->{error}->{new_num}->{$nr}++;
-				#print "ERROR $id new in $file\n";
-			}
 		}
 
 		$stat->{exists}->{$nr}++ if exists $stat->{_}->{ $id };
@@ -87,9 +78,7 @@ print "# total = ",scalar keys %{ $stat->{_} }, $/;
 foreach my $id ( keys %{ $stat->{_} } ) {
 	$stat->{count}->{ scalar @{ $stat->{_}->{$id} } }++;
 	$stat->{count_total}++;
-	foreach my $val ( @{  $stat->{_}->{$id} } ) {
-		#$stat->{val}->{$val}++ if $val > 1;
-	}
+	$stat->{count_val}->{ join(' ', @{ $stat->{_}->{$id} }) }++;
 }
 print "# stat = ",dump( $stat );
 #print "# keys = ",dump( $keys );
-- 
2.20.1